Two complementary surveys were carried out in the 89 hospital units of South-Eastern France which deal with pregnant women. Firstly, in November 1991, medical chiefs of these units were interviewed face-to-face about their current HIV screening policy. Secondly, between Jan 27 and March 22, 1992, all women at the end of their pregnancy attending these units were included in an anonymous unlinked seroprevalence survey, irrespective of pregnancy outcome (n = 11,056). The goal of the research was to compare HIV prenatal screening policies and seroprevalences by pregnancy outcomes in order to contribute to the public debate initiated on that issue by the French health authorities. The seroprevalence survey showed a global prevalence rate of 0.43% (CI 95% = 0.32-0.54) with the prevalence among women who had an elective abortion (0.56%) being more than twice that among women who delivered (0.22%). However, routine HIV screening was more frequent toward women coming for regular prenatal care than for women seeking abortion. A systematic procedure for obtaining women's consent for HIV testing only existed in a minority of units. Only 23 out of the 62 units offering both antenatal and termination services to women had the same screening policy for women attending the different services. The research confirmed that a mandatory requirement would not improve HIV screening policy during prenatal care. However, less emphasis on women who have opted for termination of pregnancy, an absence of appropriate counselling and information procedures, and pressures on HIV-infected women to terminate current pregnancies and discourage future ones strongly suggest that HIV prenatal screening in French hospitals remains mainly focussed on fetal concerns, without sufficient attention towards the needs of women at risk of HIV infection.